2 * Copyright 2006-2007 Adrian Thurston <thurston@cs.queensu.ca>
5 /* This file is part of Ragel.
7 * Ragel is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * Ragel is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with Ragel; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
28 #include "parsedata.h"
40 /* This is used for tracking the current stack of include file/machine pairs. It is
41 * is used to detect and recursive include structure. */
42 struct IncludeStackItem
44 IncludeStackItem( char *fileName, char *sectionName )
45 : fileName(fileName), sectionName(sectionName) {}
51 typedef Vector<IncludeStackItem> IncludeStack;
61 Scanner( char *fileName, istream &input, ostream &output,
62 Parser *inclToParser, char *inclSectionTarg,
65 fileName(fileName), input(input), output(output),
66 inclToParser(inclToParser),
67 inclSectionTarg(inclSectionTarg),
68 includeDepth(includeDepth),
69 line(1), column(1), lastnl(0),
70 parser(0), active(false),
71 parserExistsError(false),
75 bool recursiveInclude( char *inclFileName, char *inclSectionName );
77 char *prepareFileName( char *fileName, int len )
80 Token tokenFnStr, tokenRes;
81 tokenFnStr.data = fileName;
82 tokenFnStr.length = len;
83 tokenFnStr.prepareLitString( tokenRes, caseInsensitive );
88 void token( int type, char *start, char *end );
89 void token( int type, char c );
90 void token( int type );
96 ostream &scan_error();
101 Parser *inclToParser;
102 char *inclSectionTarg;
108 int word_len, lit_len;
110 char *tokstart, *tokend;
114 /* Set by machine statements, these persist from section to section
115 * allowing for unnamed sections. */
118 IncludeStack includeStack;
120 /* This is set if ragel has already emitted an error stating that
121 * no section name has been seen and thus no parser exists. */
122 bool parserExistsError;
124 /* This is for inline code. By default it is on. It goes off for
125 * statements and values in inline blocks which are parsed. */
130 machine section_parse;
135 void Scanner::init( )
140 bool Scanner::parserExists()
145 if ( ! parserExistsError ) {
146 scan_error() << "include: there is no previous specification name" << endl;
147 parserExistsError = true;
152 ostream &Scanner::scan_error()
154 /* Maintain the error count. */
156 cerr << fileName << ":" << line << ":" << column << ": ";
160 bool Scanner::recursiveInclude( char *inclFileName, char *inclSectionName )
162 for ( IncludeStack::Iter si = includeStack; si.lte(); si++ ) {
163 if ( strcmp( si->fileName, inclFileName ) == 0 &&
164 strcmp( si->sectionName, inclSectionName ) == 0 )
172 void Scanner::updateCol()
177 //cerr << "adding " << tokend - from << " to column" << endl;
178 column += tokend - from;
182 void Scanner::token( int type, char c )
184 token( type, &c, &c + 1 );
187 void Scanner::token( int type )
193 machine section_parse;
195 # This relies on the the kelbt implementation and the order
196 # that tokens are declared.
203 action clear_words { word = lit = 0; word_len = lit_len = 0; }
204 action store_word { word = tokdata; word_len = toklen; }
205 action store_lit { lit = tokdata; lit_len = toklen; }
207 action mach_err { scan_error() << "bad machine statement" << endl; }
208 action incl_err { scan_error() << "bad include statement" << endl; }
209 action write_err { scan_error() << "bad write statement" << endl; }
211 action handle_machine
213 /* Assign a name to the machine. */
214 char *machine = word;
216 if ( inclSectionTarg == 0 ) {
219 ParserDictEl *pdEl = parserDict.find( machine );
221 pdEl = new ParserDictEl( machine );
222 pdEl->value = new Parser( fileName, machine, sectionLoc );
224 parserDict.insert( pdEl );
227 parser = pdEl->value;
229 else if ( strcmp( inclSectionTarg, machine ) == 0 ) {
230 /* found include target */
232 parser = inclToParser;
235 /* ignoring section */
242 ( KW_Machine TK_Word @store_word ';' ) @handle_machine
243 <>err mach_err <>eof mach_err;
245 action handle_include
247 if ( active && parserExists() ) {
248 char *inclSectionName = word;
249 char *inclFileName = 0;
251 /* Implement defaults for the input file and section name. */
252 if ( inclSectionName == 0 )
253 inclSectionName = parser->sectionName;
256 inclFileName = prepareFileName( lit, lit_len );
258 inclFileName = fileName;
260 /* Check for a recursive include structure. Add the current file/section
261 * name then check if what we are including is already in the stack. */
262 includeStack.append( IncludeStackItem( fileName, parser->sectionName ) );
264 if ( recursiveInclude( inclFileName, inclSectionName ) )
265 scan_error() << "include: this is a recursive include operation" << endl;
267 /* Open the input file for reading. */
268 ifstream *inFile = new ifstream( inclFileName );
269 if ( ! inFile->is_open() ) {
270 scan_error() << "include: could not open " <<
271 inclFileName << " for reading" << endl;
274 Scanner scanner( inclFileName, *inFile, output, parser,
275 inclSectionName, includeDepth+1 );
281 /* Remove the last element (len-1) */
282 includeStack.remove( -1 );
287 TK_Word @store_word ( TK_Literal @store_lit )? |
288 TK_Literal @store_lit
292 ( KW_Include include_names ';' ) @handle_include
293 <>err incl_err <>eof incl_err;
297 if ( active && machineSpec == 0 && machineName == 0 ) {
299 " def_name=\"" << parser->sectionName << "\""
300 " line=\"" << line << "\""
301 " col=\"" << column << "\""
308 if ( active && machineSpec == 0 && machineName == 0 )
309 output << "<arg>" << tokdata << "</arg>";
314 if ( active && machineSpec == 0 && machineName == 0 )
315 output << "</write>\n";
319 ( KW_Write @write_command
320 ( TK_Word @write_arg )+ ';' @write_close )
321 <>err write_err <>eof write_err;
325 /* Send the token off to the parser. */
326 if ( active && parserExists() ) {
330 cerr << "scanner:" << line << ":" << column <<
331 ": sending token to the parser " << lelNames[*p];
332 cerr << " " << toklen;
334 cerr << " " << tokdata;
338 loc.fileName = fileName;
342 parser->token( loc, type, tokdata, toklen );
346 # Catch everything else.
347 everything_else = ^( KW_Machine | KW_Include | KW_Write ) @handle_token;
357 void Scanner::token( int type, char *start, char *end )
366 tokdata = new char[toklen+1];
367 memcpy( tokdata, start, toklen );
372 machine section_parse;
379 void Scanner::startSection( )
381 parserExistsError = false;
383 if ( includeDepth == 0 ) {
384 if ( machineSpec == 0 && machineName == 0 )
385 output << "</host>\n";
388 sectionLoc.fileName = fileName;
389 sectionLoc.line = line;
393 void Scanner::endSection( )
395 /* Execute the eof actions for the section parser. */
397 machine section_parse;
401 /* Close off the section with the parser. */
402 if ( active && parserExists() ) {
404 loc.fileName = fileName;
408 parser->token( loc, TK_EndSection, 0, 0 );
411 if ( includeDepth == 0 ) {
412 if ( machineSpec == 0 && machineName == 0 ) {
413 /* The end section may include a newline on the end, so
414 * we use the last line, which will count the newline. */
415 output << "<host line=\"" << line << "\">";
423 # This is sent by the driver code.
433 # Identifiers, numbers, commetns, and other common things.
434 ident = ( alpha | '_' ) ( alpha |digit |'_' )*;
436 hex_number = '0x' [0-9a-fA-F]+;
439 '/*' ( any | NL )* :>> '*/';
444 c_cpp_comment = c_comment | cpp_comment;
446 # These literal forms are common to C-like host code and ragel.
447 s_literal = "'" ([^'\\] | NL | '\\' (any | NL))* "'";
448 d_literal = '"' ([^"\\] | NL | '\\' (any | NL))* '"';
450 whitespace = [ \t] | NL;
451 pound_comment = '#' [^\n]* NL;
453 # An inline block of code. This is specified as a scanned, but is sent to
454 # the parser as one long block. The inline_block pointer is used to handle
455 # the preservation of the data.
457 # Inline expression keywords.
458 "fpc" => { token( KW_PChar ); };
459 "fc" => { token( KW_Char ); };
460 "fcurs" => { token( KW_CurState ); };
461 "ftargs" => { token( KW_TargState ); };
463 whitespaceOn = false;
467 # Inline statement keywords.
469 whitespaceOn = false;
472 "fexec" => { token( KW_Exec, 0, 0 ); };
474 whitespaceOn = false;
478 whitespaceOn = false;
482 whitespaceOn = false;
486 whitespaceOn = false;
490 whitespaceOn = false;
494 ident => { token( TK_Word, tokstart, tokend ); };
496 number => { token( TK_UInt, tokstart, tokend ); };
497 hex_number => { token( TK_Hex, tokstart, tokend ); };
499 ( s_literal | d_literal )
500 => { token( IL_Literal, tokstart, tokend ); };
504 token( IL_WhiteSpace, tokstart, tokend );
506 c_cpp_comment => { token( IL_Comment, tokstart, tokend ); };
508 "::" => { token( TK_NameSep, tokstart, tokend ); };
510 # Some symbols need to go to the parser as with their cardinal value as
511 # the token type (as opposed to being sent as anonymous symbols)
512 # because they are part of the sequences which we interpret. The * ) ;
513 # symbols cause whitespace parsing to come back on. This gets turned
514 # off by some keywords.
518 token( *tokstart, tokstart, tokend );
519 if ( inlineBlockType == SemiTerminated )
525 token( *tokstart, tokstart, tokend );
528 [,(] => { token( *tokstart, tokstart, tokend ); };
531 token( IL_Symbol, tokstart, tokend );
536 if ( --curly_count == 0 && inlineBlockType == CurlyDelimited ) {
537 /* Inline code block ends. */
542 /* Either a semi terminated inline block or only the closing
543 * brace of some inner scope, not the block's closing brace. */
544 token( IL_Symbol, tokstart, tokend );
549 scan_error() << "unterminated code block" << endl;
552 # Send every other character as a symbol.
553 any => { token( IL_Symbol, tokstart, tokend ); };
557 # Escape sequences in OR expressions.
558 '\\0' => { token( RE_Char, '\0' ); };
559 '\\a' => { token( RE_Char, '\a' ); };
560 '\\b' => { token( RE_Char, '\b' ); };
561 '\\t' => { token( RE_Char, '\t' ); };
562 '\\n' => { token( RE_Char, '\n' ); };
563 '\\v' => { token( RE_Char, '\v' ); };
564 '\\f' => { token( RE_Char, '\f' ); };
565 '\\r' => { token( RE_Char, '\r' ); };
566 '\\\n' => { updateCol(); };
567 '\\' any => { token( RE_Char, tokstart+1, tokend ); };
569 # Range dash in an OR expression.
570 '-' => { token( RE_Dash, 0, 0 ); };
572 # Terminate an OR expression.
573 ']' => { token( RE_SqClose ); fret; };
576 scan_error() << "unterminated OR literal" << endl;
579 # Characters in an OR expression.
580 [^\]] => { token( RE_Char, tokstart, tokend ); };
585 # Escape sequences in regular expressions.
586 '\\0' => { token( RE_Char, '\0' ); };
587 '\\a' => { token( RE_Char, '\a' ); };
588 '\\b' => { token( RE_Char, '\b' ); };
589 '\\t' => { token( RE_Char, '\t' ); };
590 '\\n' => { token( RE_Char, '\n' ); };
591 '\\v' => { token( RE_Char, '\v' ); };
592 '\\f' => { token( RE_Char, '\f' ); };
593 '\\r' => { token( RE_Char, '\r' ); };
594 '\\\n' => { updateCol(); };
595 '\\' any => { token( RE_Char, tokstart+1, tokend ); };
597 # Terminate an OR expression.
599 token( RE_Slash, tokstart, tokend );
603 # Special characters.
604 '.' => { token( RE_Dot ); };
605 '*' => { token( RE_Star ); };
607 '[' => { token( RE_SqOpen ); fcall or_literal; };
608 '[^' => { token( RE_SqOpenNeg ); fcall or_literal; };
611 scan_error() << "unterminated regular expression" << endl;
614 # Characters in an OR expression.
615 [^\/] => { token( RE_Char, tokstart, tokend ); };
618 # We need a separate token space here to avoid the ragel keywords.
619 write_statement := |*
620 ident => { token( TK_Word, tokstart, tokend ); } ;
621 [ \t\n]+ => { updateCol(); };
622 ';' => { token( ';' ); fgoto parser_def; };
625 scan_error() << "unterminated write statement" << endl;
629 # Parser definitions.
631 'machine' => { token( KW_Machine ); };
632 'include' => { token( KW_Include ); };
635 fgoto write_statement;
637 'action' => { token( KW_Action ); };
638 'alphtype' => { token( KW_AlphType ); };
640 # FIXME: Enable this post 5.17.
641 # 'range' => { token( KW_Range ); };
645 inlineBlockType = SemiTerminated;
650 inlineBlockType = SemiTerminated;
654 token( KW_Variable );
655 inlineBlockType = SemiTerminated;
658 'when' => { token( KW_When ); };
659 'eof' => { token( KW_Eof ); };
660 'err' => { token( KW_Err ); };
661 'lerr' => { token( KW_Lerr ); };
662 'to' => { token( KW_To ); };
663 'from' => { token( KW_From ); };
666 ident => { token( TK_Word, tokstart, tokend ); } ;
669 number => { token( TK_UInt, tokstart, tokend ); };
670 hex_number => { token( TK_Hex, tokstart, tokend ); };
672 # Literals, with optionals.
673 ( s_literal | d_literal ) [i]?
674 => { token( TK_Literal, tokstart, tokend ); };
676 '[' => { token( RE_SqOpen ); fcall or_literal; };
677 '[^' => { token( RE_SqOpenNeg ); fcall or_literal; };
679 '/' => { token( RE_Slash ); fgoto re_literal; };
682 pound_comment => { updateCol(); };
684 ':=' => { token( TK_ColonEquals ); };
687 ">~" => { token( TK_StartToState ); };
688 "$~" => { token( TK_AllToState ); };
689 "%~" => { token( TK_FinalToState ); };
690 "<~" => { token( TK_NotStartToState ); };
691 "@~" => { token( TK_NotFinalToState ); };
692 "<>~" => { token( TK_MiddleToState ); };
695 ">*" => { token( TK_StartFromState ); };
696 "$*" => { token( TK_AllFromState ); };
697 "%*" => { token( TK_FinalFromState ); };
698 "<*" => { token( TK_NotStartFromState ); };
699 "@*" => { token( TK_NotFinalFromState ); };
700 "<>*" => { token( TK_MiddleFromState ); };
703 ">/" => { token( TK_StartEOF ); };
704 "$/" => { token( TK_AllEOF ); };
705 "%/" => { token( TK_FinalEOF ); };
706 "</" => { token( TK_NotStartEOF ); };
707 "@/" => { token( TK_NotFinalEOF ); };
708 "<>/" => { token( TK_MiddleEOF ); };
710 # Global Error actions.
711 ">!" => { token( TK_StartGblError ); };
712 "$!" => { token( TK_AllGblError ); };
713 "%!" => { token( TK_FinalGblError ); };
714 "<!" => { token( TK_NotStartGblError ); };
715 "@!" => { token( TK_NotFinalGblError ); };
716 "<>!" => { token( TK_MiddleGblError ); };
718 # Local error actions.
719 ">^" => { token( TK_StartLocalError ); };
720 "$^" => { token( TK_AllLocalError ); };
721 "%^" => { token( TK_FinalLocalError ); };
722 "<^" => { token( TK_NotStartLocalError ); };
723 "@^" => { token( TK_NotFinalLocalError ); };
724 "<>^" => { token( TK_MiddleLocalError ); };
727 "<>" => { token( TK_Middle ); };
730 '>?' => { token( TK_StartCond ); };
731 '$?' => { token( TK_AllCond ); };
732 '%?' => { token( TK_LeavingCond ); };
734 '..' => { token( TK_DotDot ); };
735 '**' => { token( TK_StarStar ); };
736 '--' => { token( TK_DashDash ); };
737 '->' => { token( TK_Arrow ); };
738 '=>' => { token( TK_DoubleArrow ); };
740 ":>" => { token( TK_ColonGt ); };
741 ":>>" => { token( TK_ColonGtGt ); };
742 "<:" => { token( TK_LtColon ); };
744 # Opening of longest match.
745 "|*" => { token( TK_BarStar ); };
753 [ \t]+ => { updateCol(); };
755 # If we are in a single line machine then newline may end the spec.
758 if ( singleLineSpec ) {
767 inlineBlockType = CurlyDelimited;
772 scan_error() << "unterminated ragel section" << endl;
775 any => { token( *tokstart ); } ;
781 /* If no errors and we are at the bottom of the include stack (the
782 * source file listed on the command line) then write out the data. */
783 if ( includeDepth == 0 && machineSpec == 0 && machineName == 0 )
784 xmlEscapeHost( output, tokstart, tokend-tokstart );
787 # Outside code scanner. These tokens get passed through.
791 c_cpp_comment => pass;
792 s_literal | d_literal => pass;
795 singleLineSpec = false;
801 singleLineSpec = true;
814 void Scanner::do_scan()
817 char *buf = new char[bufsize];
818 const char last_char = 0;
819 int cs, act, have = 0;
823 bool singleLineSpec = false;
824 InlineBlockType inlineBlockType;
829 char *p = buf + have;
830 int space = bufsize - have;
833 /* We filled up the buffer trying to scan a token. Grow it. */
834 bufsize = bufsize * 2;
835 char *newbuf = new char[bufsize];
837 /* Recompute p and space. */
839 space = bufsize - have;
841 /* Patch up pointers possibly in use. */
843 tokstart = newbuf + ( tokstart - buf );
844 tokend = newbuf + ( tokend - buf );
846 /* Copy the new buffer in. */
847 memcpy( newbuf, buf, have );
852 input.read( p, space );
853 int len = input.gcount();
855 /* If we see eof then append the EOF char. */
857 p[0] = last_char, len = 1;
864 /* Check if we failed. */
865 if ( cs == rlscan_error ) {
866 /* Machine failed before finding a token. I'm not yet sure if this
868 scan_error() << "scanner error" << endl;
872 /* Decide if we need to preserve anything. */
873 char *preserve = tokstart;
875 /* Now set up the prefix. */
879 /* There is data that needs to be shifted over. */
880 have = pe - preserve;
881 memmove( buf, preserve, have );
882 unsigned int shiftback = preserve - buf;
884 tokstart -= shiftback;
894 void scan( char *fileName, istream &input, ostream &output )
896 Scanner scanner( fileName, input, output, 0, 0, 0 );
901 eofLoc.fileName = fileName;
903 eofLoc.line = scanner.line;